Inducing causal relationships from observations is a classic problem in scientific inference, statistics, and machine learning. It is also a central part of human learning, and a task that people perform remarkably well given its notorious difficulties. People can learn causal structure in various settings, from diverse forms of data: observations of the co-occurrence frequencies between causes and effects, interactions between physical objects, or patterns of spatial or temporal coincidence. These different modes of learning are typically thought of as distinct psychological processes and are rarely studied together, but at heart they present the same inductive challenge-identifying the unobservable mechanisms that generate observable relations between variables, objects, or events, given only sparse and limited data. We present a computational-level analysis of this inductive problem and a framework for its solution, which allows us to model all these forms of causal learning in a common language. In this framework, causal induction is the product of domain-general statistical inference guided by domain-specific prior knowledge, in the form of an abstract causal theory. We identify 3 key aspects of abstract prior knowledge-the ontology of entities, properties, and relations that organizes a domain; the plausibility of specific causal relationships; and the functional form of those relationships-and show how they provide the constraints that people need to induce useful causal models from sparse data.
Keywords: causal induction, intuitive theories, rational analysis, Bayesian modelingIn 1695, Sir Edmond Halley was computing the orbits of a set of comets for inclusion in Newton's Principia Mathematica when he noticed a surprising regularity: The comets of 1531, 1607, and 1682 took remarkably similar paths across the sky, and visited the Earth approximately 76 years apart. Newton had already shown that comets should follow orbits corresponding to conic sectionsparabolas, hyperbolas, and ellipses-although no elliptical orbits had yet been observed. Halley inferred that the sightings of these comets were not three independent events, but three consequences of a single common cause: a comet that had visited the Earth three times, travelling in an elliptical orbit. He went on to predict that it would return along the same orbit in 1758. The comet returned as predicted, and has continued to visit the Earth approximately every 76 years since, providing a sensational confirmation of Newton's physics.Halley's discovery is an example of causal induction: inferring causal structure from data. Explaining this discovery requires appealing to two factors: abstract prior knowledge, in the form of a causal theory, and statistical inference. The prior knowledge that guided Halley was the mathematical theory of physics laid out by Newton. This theory identified the entities and properties relevant to understanding a physical system, formalizing notions such as velocity and acceleration, and characterized the relations that can ho...